In a digital communication system, as a communication rate increases continuously, phase noise already becomes one of important factors that affect performance of a high-speed digital communication system. With continuous development of radio technologies, application of a multi-path system, such as a MIMO (multiple-input multiple-output) and adaptive array receiving system, receives increasingly extensive attention and study and is a technology development trend in the future. At present, however, when signals in a multi-path system are corrected, a same signal correction method is simply and repeatedly used on each channel. When a used signal correction method is based on adaptive filtering, phase noise is estimated or predicted generally through training. A decision feedback manner is used in an unknown data phase, and change of a channel may cause a decision error and further affect an adaptive algorithm, leading to an output error of a part of channels among multiple channels and making subsequent processing for multiple channels of signals greatly affected.